Learning to View: Decision Transformers for Active Object Detection

TL;DR

This paper introduces a reinforcement learning approach using Decision Transformers for active object detection, enabling robots to plan views that improve detection quality through online fine-tuning.

Contribution

It presents a novel application of Decision Transformers with online fine-tuning for active perception, enhancing detection performance in robotic systems.

Findings

Outperforms baseline methods including expert policies and offline RL.

Demonstrates improved detection quality through active view planning.

Provides detailed analysis of reward and observation spaces.

Abstract

Active perception describes a broad class of techniques that couple planning and perception systems to move the robot in a way to give the robot more information about the environment. In most robotic systems, perception is typically independent of motion planning. For example, traditional object detection is passive: it operates only on the images it receives. However, we have a chance to improve the results if we allow planning to consume detection signals and move the robot to collect views that maximize the quality of the results. In this paper, we use reinforcement learning (RL) methods to control the robot in order to obtain images that maximize the detection quality. Specifically, we propose using a Decision Transformer with online fine-tuning, which first optimizes the policy with a pre-collected expert dataset and then improves the learned policy by exploring better solutions in the environment. We evaluate the performance of proposed method on an interactive dataset collected from an indoor scenario simulator. Experimental results demonstrate that our method outperforms all baselines, including expert policy and pure offline RL methods. We also provide exhaustive analyses of the reward distribution and observation space.